Soumettre #553644: pytorch pytorch (in torch.cuda.nccl.reduce) 2.6.0 Denial of Service

When using `torch.cuda.nccl.reduce` with invalid operation codes, the program crashes with `Aborted (core dumped)` instead of raising a `RuntimeError` or validating the input. This bug can be exploited for a denial-of-service attack.

# To Reproduce
```
import torch
import torch.cuda as cuda
from torch.cuda.nccl import reduce
def test_bug():
# Checking for unsupported operations
unsupported_ops = [0xFF, 0xAA] # Example of invalid NCCL operation codes
for op in unsupported_ops:
input_tensor = torch.tensor([1.0, 2.0, 3.0], device=f'cuda:0')
output_tensor = torch.zeros_like(input_tensor)
reduce(inputs=[input_tensor], output=output_tensor, root=0, op=op)
if __name__ == "__main__":
test_bug()
```

# Output
```
Aborted (core dumped)
```

• La source: ⚠️ https://github.com/pytorch/pytorch/issues/150836
• Utilisateur: Default436352 (UID 81891)
• Soumission: 08/04/2025 17:11 (il y a 12 mois)
• Modérer: 05/05/2025 13:53 (27 days later)
• Statut: Accepté
• Entrée VulDB: 307394 [PyTorch 2.6.0+cu124 torch/cuda/nccl.py torch.cuda.nccl.reduce déni de service]
• Points: 20